DEUNet: Dual-encoder UNet for simultaneous denoising and reconstruction of single HDR image

被引:4
作者
Yao, Zhiping [1 ]
Bi, Jiang [2 ]
Deng, Wei [3 ]
He, Wenlin [2 ]
Wang, Zihan [1 ]
Kuang, Xu [1 ]
Zhou, Mi [4 ]
Gao, Qinquan [1 ]
Tong, Tong [1 ]
机构
[1] Fuzhou Univ, Coll Phys & Informat Engn, Fuzhou, Peoples R China
[2] Beijing Radio & Televis Stn, Beijing, Peoples R China
[3] Imperial Vis Technol, Fujian, Peoples R China
[4] Beijing Tongzhou Media Convergence Ctr, Beijing, Peoples R China
来源
COMPUTERS & GRAPHICS-UK | 2024年 / 119卷
基金
中国国家自然科学基金;
关键词
HDR reconstruction; Deep learning; Image denoising; Image enhancement; DYNAMIC-RANGE EXPANSION; FILTERS;
D O I
10.1016/j.cag.2024.01.009
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
High Dynamic Range (HDR) images have more powerful image expression capabilities and better visual quality than Low Dynamic Range (LDR) images, which can sufficiently represent real -world scenes, and will be widely used in the field of film and television in future. However, it is a very challenging task to generate an HDR image from a single -exposure LDR image. In this work, we propose a novel learning -based network, DEUNet, to reconstruct single -frame HDR image with simultaneous denoising and detail reconstruction. The proposed framework consists of two feature extraction branches, which can learn the brightness information and texture information separately for HDR image reconstruction. Each network branch is based on the UNet network structure and the two branches are interacted via spatial feature transformation. As a result, the proposed network can make full use of the multi -scale information at different levels of the image. In addition to the two encoding branches for feature extraction, the proposed network consists of another decoding network for fusing image brightness information and texture information, and a weighting network that selectively preserves most useful information. Compared with state-of-the-art methods, DEUNet can better reduce image noise while reconstructing the details in both the high and the low exposure areas. Experiments have shown that the proposed method achieves state-of-the-art performance on public datasets, indicating the effectiveness of the proposed method in this study.
引用
收藏
页数:10
相关论文
共 34 条
  • [21] A Decoupled Kernel Prediction Network Guided by Soft Mask for Single Image HDR Reconstruction
    Cao, Gaofeng
    Zhou, Fei
    Liu, Kanglin
    Wang, Anjie
    Fan, Leidong
    ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS, 2023, 19 (02)
  • [22] Use of Local Region Maps on Convolutional LSTM for Single-Image HDR Reconstruction
    Oh, Seungwook
    Shin, GyeongIk
    Hong, Hyunki
    CMC-COMPUTERS MATERIALS & CONTINUA, 2022, 71 (03): : 4555 - 4572
  • [23] MWIR image deep denoising reconstruction based on single-pixel imaging
    Yang, Shuowen
    Qin, Hanlin
    Yan, Xiang
    Zhao, Dong
    Zeng, Qingjie
    OPTICS COMMUNICATIONS, 2025, 574
  • [24] Low-dose CT image denoising via frequency division and encoder-dual decoder GAN
    Jiao, Fengyuan
    Gui, Zhiguo
    Liu, Yi
    Yao, Linhong
    Zhang, Pengcheng
    SIGNAL IMAGE AND VIDEO PROCESSING, 2021, 15 (08) : 1907 - 1915
  • [25] Low-dose CT image denoising via frequency division and encoder-dual decoder GAN
    Fengyuan Jiao
    Zhiguo Gui
    Yi Liu
    Linhong Yao
    Pengcheng Zhang
    Signal, Image and Video Processing, 2021, 15 : 1907 - 1915
  • [26] Single image denoising based on adaptive fusion dual-domain network
    Jiang, Zhiwei
    Xue, Zhizhong
    Wang, Jue
    Hu, Yibiao
    Zheng, Qiufu
    IET IMAGE PROCESSING, 2024, 18 (03) : 561 - 571
  • [27] Evaluation of Image Quality and Detectability of Deep Learning Image Reconstruction (DLIR) Algorithm in Single- and Dual-energy CT
    Jingyu Zhong
    Hailin Shen
    Yong Chen
    Yihan Xia
    Xiaomeng Shi
    Wei Lu
    Jianying Li
    Yue Xing
    Yangfan Hu
    Xiang Ge
    Defang Ding
    Zhenming Jiang
    Weiwu Yao
    Journal of Digital Imaging, 2023, 36 : 1390 - 1407
  • [28] Evaluation of Image Quality and Detectability of Deep Learning Image Reconstruction (DLIR) Algorithm in Single- and Dual-energy CT
    Zhong, Jingyu
    Shen, Hailin
    Chen, Yong
    Xia, Yihan
    Shi, Xiaomeng
    Lu, Wei
    Li, Jianying
    Xing, Yue
    Hu, Yangfan
    Ge, Xiang
    Ding, Defang
    Jiang, Zhenming
    Yao, Weiwu
    JOURNAL OF DIGITAL IMAGING, 2023, 36 (04) : 1390 - 1407
  • [29] Fully Convolutional Denoising Autoencoder for 3D Scene Reconstruction from a single depth image
    Palla, Alessandro
    Moloney, David
    Fanucci, Luca
    2017 4TH INTERNATIONAL CONFERENCE ON SYSTEMS AND INFORMATICS (ICSAI), 2017, : 566 - 575
  • [30] Single Exposure High Dynamic Range Image Reconstruction Based on Deep Dual-Branch Network
    Ye, Nianjin
    Huo, Yongqing
    Liu, Shuaicheng
    Li, Hanlin
    IEEE ACCESS, 2021, 9 : 9610 - 9624